The Capability Set
The capability set is not a list of things a person does. It is the full range of things a person could do — the set of genuinely achievable functionings from which she is free to choose. The value of the capability set lies not in its exercise but in its existence. A person who could engage in creative work but chooses rest has a larger capability set than a person who rests because creative work is unavailable. The freedom inheres in the choice, not in the outcome. This concept is conspicuously absent from the AI discourse, which tends to celebrate every expansion of functioning as if it were an expansion of capability.
In the AI Story
The capability set concept produces evaluative precision that output metrics cannot match. Consider two engineers using the same AI tool to achieve the same twenty-fold productivity multiplier. In the output space, they are identical. In the capability set space, they may be radically different. The first uses the freed cognitive bandwidth to move into product strategy, exercising judgment she did not previously exercise — her capability set has expanded to include strategic thinking. The second uses the tool to take on more tasks, filling every freed minute with additional execution — her capability set has not expanded; only her output has.
The distinction matters particularly in contexts where preferences adapt to constraints. A person whose capability set contracts through disuse — who stops debugging manually, loses architectural intuition, ceases to value the capacities she once possessed — may report satisfaction with AI-augmented work. The satisfaction is real. The adaptive preference mechanism, however, renders the contraction invisible to any evaluative framework that takes satisfaction as its measure.
Sen's framework treats the capability set as the proper unit of ethical analysis because it captures what matters about freedom in a way that functionings alone cannot. A person who has only one available functioning has no freedom, even if she performs the functioning well. A person with a large capability set has freedom even if she exercises only a small subset of the available functionings. The size and quality of the set, not the selected functionings, are the measure of freedom.
Applied to AI, the capability set concept reframes the central question. The question is not whether AI makes people more productive. It is whether AI expands or contracts the range of genuinely available choices about how to work, what to create, and what kind of life to build around the work. The answer depends on conversion factors that the technology itself does not control and the technology industry does not measure.
Origin
The formal mathematical representation of the capability set appears in Sen's Commodities and Capabilities (1985), where he developed the vector representation of functionings and the set-theoretic framework for aggregating them into capabilities.
Key Ideas
Range over selection. What matters is the range of achievable functionings, not the particular functionings selected.
Freedom inheres in unexercised options. The person who could do something but chooses not to is freer than the person who cannot do it.
Expansion versus substitution. AI can expand functionings while contracting capabilities if it substitutes for capacities it does not develop.
The set is contextual. The same tool produces different capability sets in different institutional environments.
Appears in the Orange Pill Cycle
Further reading
- Amartya Sen, Commodities and Capabilities (North-Holland, 1985)
- Sabina Alkire, Valuing Freedoms: Sen's Capability Approach and Poverty Reduction (Oxford University Press, 2002)